Korean Journal of Remote Sensing, Vol.33, No.1, 2017, pp.37~46 ISSN 1225-6161 ( Print ) http://dx.doi.org/10.7780/kjrs.2017.33.1.4 ISSN 2287-9307 (Online)

Article

Monitoring Mount Sinabung in Using Multi- Temporal InSAR

Chang-Wook Lee*, Zhong Lu** and Jin Woo Kim** † * Division of Science Education, College of Education, Kangwon National University ** Roy M Huffington Department of Earth Sciences, Southern Methodist University

Sinabung volcano in Indonesia was formed due to the subduction between the Eurasian and Abstract : Indo-Australian plates along the Pacific . After being dormant for about 400 years, Sinabung volcano erupted on the 29th of August, 2010 and most recently on the 1st of November, 2016. We measured the deformation of Sinabung volcano using Advanced Land Observing Satellite/Phased Array type L-band Synthetic Aperture Radar (ALOS/PALSAR) interferometric synthetic aperture radar (InSAR) images acquired from February 2007 to January 2011. Based on multi-temporal InSAR processing, we mapped the ground surface deformation before, during, and after the 2010 eruption with time-series InSAR technique. During the 3 years before the 2010 eruption, the volcano inflated at an average rate of ~1.7 cm/yr with a markedly higher rate of 6.6 cm/yr during the 6 months prior to the 2010 eruption. The inflation was constrained to the top of the volcano. From the 2010 eruption to January 2011, the volcano subsided by approximately 3 cm (~6 cm/yr). We interpreted that the inflation was due to magma accumulation in a shallow reservoir beneath Sinabung. The deflation was attributed to magma withdrawal from the shallow reservoir during the eruption as well as thermo-elastic compaction of erupted material. This result demonstrates once again the utility of InSAR for volcano monitoring. Sinabung volcano, surface deformation, InSAR, time-series Key Words :

1. Introduction significant seismic and volcanic activity. The Island of has been the location for three of the world’s The Island of Sumatra is located along a segment of twelve largest earthquakes all with magnitudes higher the “Ring of Fire”, on the northern boundary of the than Mw 8.5. The 2004 Sumatra earthquake, which is Australian plate and the southern boundary of the the third largest earthquake (Mw 9.1), was particularly Eurasian plate (Global Volcanism Program). From the devastating because it was accompanied by a tsunami geologic record it is apparent that during the past 2.5 that killed more than 227,000 people from 14 countries million years, this region has been experiencing in South Asia and East Africa. The largest eruption

Received February 9, 2017; Revised February 10, 2017; Accepted February 11, 2017. † Corresponding Author: Jin Woo Kim ([email protected]) This is an Open-Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons. org/licenses/by-nc/3.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited

–37 – Korean Journal of Remote Sensing , Vol .33, No .1, 2017 event in this region, with a volcanic explosivity index volcano has been poorly monitored. The latest eruption (VEI) of 8, occurred between 69,000 and 77,000 years at Sinabung began on the 29th of August, 2010. This ago at Toba volcano on the Island of Sumatra, less than explosive eruption produced ash plumes rising several 50 km southeast of Sinabung. kilometers above the crater and forced the evacuation Mount Sinabung is a in the of 20,000 to 30,000 people within a 6 km radius of the northwestern part of the Island of Sumatra in Indonesia volcano. Two individual ash plumes from separate vents (Fig. 1a). The volcano was formed during the near the summit were produced during this eruption, one Pleistocene to Holocene and consists of andesitic and vertical and one horizontal (Figs. 2a and b). Explosions dacitic . Sinabung’s 2,460-m-high conical structure and ash plumes accompanied by strong volcanic has four overlapping craters (Global Volcanism tremors, which were felt by residents as far as 8 km Program). It is mostly covered with dense vegetation away, continued until the 22nd of September. and only the top of the volcano is free of vegetation (Fig. According to the Center of Volcanology and Geological 1b). The four summit vents are elongated from north- Hazard Mitigation (CVGHM), the ash plume, over the south and range from 60 m to 300 m in diameter (Figs. duration of this eruptio n, reached a maximum height of 1c and d). Little is known about the volcano’s eruption 5 km above the crater floor on the 7th of September. history, but the last historical eruption occurred around The volcano remained dormant for another three 400 years ago. Because of this long dormancy, the years before erupting again on the 15th of September,

Fig. 1. (a) Sinabung volcano, located in the northern region of the Island of Sumatra, Indonesia. (b) Landsat-7 image acquired on May 19, 2003 of the Mount Sinabung area. (c) Averaged SAR amplitude image between January 1, 2007 and January 16, 2011. (d) Combined Landsat-7 and averaged SAR images. Red arrows indicate the location of multi-craters and the yellow dotted lines delineate the extent of the rocky area.

–38 – Monitoring Mount Sinabung in Indonesia Using Multi-Temporal InSAR

Fig. 2. Twin ash plumes from Sinabung volcano eruption of 29-30th of August, 2010. Picture by the Jakarta Post on the 29th of August (a) and the 30th of August (b). (c) Landsat 7 image (RGB = 3,2,1) acquired on September 11, 2013.

2013. During this eruption the ash plume rose to an Simons, 2004; Lu et al ., 2010; Lu and Dzurisin, 2014; altitude of 6.1 km, and 3,700 people within a 3-km Chaussard and Amelung, 2012; Ebmeier et al ., 2010; radius of the volcano were evacuated. Moreover, Philibosian and Simon, 2011; Furuta, 2009). Because volcanic ash fell as far away as 50 km from the volcano Sinabung is heavily vegetated, there is too much loss in the northeast direction. Landsat 7 satellite image of coherence in C-band InSAR imagery. However, (Fig. 2c) acquired 4 days before the eruption Chaussard and Amelung (2010, 2012) processed (September 11, 2013) shows a white ash cloud over the thirteen L-band ALOS/PALSAR images from summit of Sinabung and a grey volcanic ash plume on Sinabung, acquired before 2010 that show a the east side of the volcano. progressive rate of inflation of 2.2 cm/yr. In this study, The size of Sinabung volcano and the frequent and we used twenty ALOS/PALSAR data (Table 1) with complex nature of its eruptions make it difficult to study calibrated radar backscatter at HH polarization, the deformation that is occurring here using traditional including seven new images acquired in 2010 and methods that require surface based instruments. InSAR January 2011, to study deformation of the volcano. is an efficient method for measuring surface We applied a small baseline subset (SBAS) InSAR deformation over large areas (Massonnet and Feigl, technique (Berardino et al ., 2002; Lee et al ., 2011) to 1998; Rignot, 2008; Ng et al ., 2009; Lu et al ., 2009; estimate time-series surface deformation using a multi- Jung et al ., 2013a), and has been applied worldwide in interferogram processing approach (Rignot, 2008; Pan the study of volcanic deformation (Pritchard and and Tang, 2010). Atmospheric artifacts are reduced

–39 – Korean Journal of Remote Sensing , Vol .33, No .1, 2017

Table 1. Characteristics of ALOS/PALSAR data used in this study Date Perpendicular Number Mission (Polarization) Orbit number (YYYYMMDD) baseline (m) 1 ALOS/PALSAR(HH) 00702 20070220 0 (M) 2 ALOS/PALSAR(HH) 00707 20070708 181(S) 3 ALOS/PALSAR(HH) 00708 20070823 205(S) 4 ALOS/PALSAR(HH) 00801 20080108 15(S) 5 ALOS/PALSAR(HH) 00802 20080223 194(S) 6 ALOS/PALSAR(HH) 00804 20080409 305(S) 7 ALOS/PALSAR(HH) 00805 20080525 -45(S) 8 ALOS/PALSAR(HH) 00810 20081010 -124(S) 9 ALOS/PALSAR(HH) 00901 20090110 101(S) 10 ALOS/PALSAR(HH) 00902 20090225 -227(S) 11 ALOS/PALSAR(HH) 00907 20090713 449(S) 12 ALOS/PALSAR(HH) 00908 20090828 546(S) 13 ALOS/PALSAR(HH) 00911 20091128 147(S) 14 ALOS/PALSAR(HH) 01001 20100113 -3(S) 15 ALOS/PALSAR(HH) 01002 20100228 189(S) 16 ALOS/PALSAR(HH) 01007 20100716 46(S) 17 ALOS/PALSAR(HH) 01008 20100831 202(S) 18 ALOS/PALSAR(HH) 01010 20101016 154(S) 19 ALOS/PALSAR(HH) 01012 20101201 -165(S) 20 ALOS/PALSAR(HH) 01101 20110116 31(S) •Perpendicular baseline (m): calculation perpendicular baseline information of Master (M) image corresponding to Slave (S) images respectively. through spatio-temporal filtering after the mean dataset, and then carried out the SBAS processing and deformation is obtained (Jung et al ., 2013b) and stacking to characterize the surface deformation of the iterative processing can further improve the time-series Sinabung volcano in spatio-temporal domain. Using deformation estimates (Lee et al ., 2011). When using twenty ALOS/PALSAR images from the 20th of this algorithm, the deformation for a specific time February, 2007 to the 16th of January, 2011 (Table 1), period can be calculated more precisely by minimizing we generated 39 interferograms (Fig. 3). all error components (Lee et al ., 2012). This technique Most of the interferograms had a perpendicular of a refined SBAS time-series analysis can successfully baseline of less than 400 m, but one interferogram had measure deformation with an error of 0.2 to 0.9 mm in a baseline of 550 m. The best coherence was the temporal domain and a root mean square error constrained on the volcano summit. Outside the (RMSE) of 0.66 mm/yr in the spatial domain (Lee et summit, coherence was maintained only at many al ., 2012). scattered patches of rocky and less-vegetated areas (i.e. selected interferograms in Fig. 4). Figs. 4(a)-(k) show interferograms acquired from the 20th of February, 2. Data processing 2007 to the 31st of August, 2010, spanning 1 to 3 years. These images show 2 to 6 cm of surface uplift in the We initially applied InSAR processing to our SAR direction of the satellite’s line of site (LOS) over the

–40 – Monitoring Mount Sinabung in Indonesia Using Multi-Temporal InSAR

Fig. 3. Thirty-nine high-coherence interferograms according to perpendicular baseline using 20 SAR images.

Fig. 4. (a)-(k) represent surface deformation (inflation) on the summit with a high-coherence area before the eruption. (l) shows surface deformation (deflation) in the opposite direction compared to (a)-(k).

–41 – Korean Journal of Remote Sensing , Vol .33, No .1, 2017 volcano summit while the interferogram from the 31st volcano summit. The time-series surface deformation of August, 2010, to the 16th of January, 2011, that generated by SBAS processing includes two noticeable spans the 2010 eruption, show 3 cm of subsidence over fluctuations around early and late 2008 at the summit the summit area. (P(1) in Fig. 6) compared to the stability at other selected points (P(2), P(3), and P(4) in Fig. 6). This phenomenon may have been caused by the 3. Results depressurization of the magma chamber due to periodic volcanic degassing at the summit of Sinabung. In An averaged surface deformation rate map (Fig. 5) accordance with this analysis, surface deformation of was generated from 39 interferograms using SBAS the summit area steadily increased from the period of processing. The areas of each interferogram with the Fig. 7(a) until the period of Fig. 7(o) with a ground coherence less than 0.4 were excluded from our uplift pattern. However, Figs. 7(p) and (q) show rapid analysis. The volcano summit (P1) experienced about ground uplift after 138 and 184 days, respectively, from 9 cm of inflation from the 20th of February, 2007, to Fig. 7(o). It is especially rapid when compared with the 31st of August, 2010 (Fig. 5). The inflation was Fig. 7(o) at the summit, where there is one fringe (4 likely due to the intrusion of magma from a deep source cm) of increased surface deformation. The deformation to the shallow reservoir. At locations P2, P3, and P4, in these interferograms is consistent with the ground surface deformation was not significant during the movement at P(1) in Fig. 5 between the 13th of observation period (Fig. 5). January, 2010, and the 31st of August, 2010. On the This suggests that the inflation at Sinabung was contrary, Figs. 7(r)-(t) displays clear deformation of one constrained within a radius of about 500 m from the fringe (4 cm) in the opposite direction, away from the

Fig. 5. SBAS-estimated mean surface deformation rate map over high coherence areas (>0.4) exploiting the SAR dataset from 2/20/07 to 1/16/11. P1 represents one of highest deformation areas in the summit. P(2-4) were selected for comparison with a major deformation area (P1). P2 is an additional point separated from the summit area of Sinabung mountain. P(3-4) are far from the summit point P1 within the Sinabung mountain area.

–42 – Monitoring Mount Sinabung in Indonesia Using Multi-Temporal InSAR

Fig. 6. Time series surface deformation estimated by SBAS processing from 2007 to 2011. Red diamond P(1) shows large surface deformation point on the summit area at Sinabung volcano. Black inverted triangle P(2) represents a reference point in the bare soil zone. Green cross P(3) and purple triangle P(4) display the northern and southern parts of the mountain from Fig. 5.

Fig. 7. Cumulative surface deformation maps around summit area (red rectangle in Fig. 5) from the SBAS processing.

–43 – Korean Journal of Remote Sensing , Vol .33, No .1, 2017 sensor, after the August-September 2010 eruption March 24 th 2011 to verify the movement of the surface events. Such a large deformation could be due to a inflation or deflation with another SAR satellite mission combined effects of depressurization of the shallow in future work because GPS stations were installed and magma reservoir due to cooling, and thermo-elastic monitored on Sinabung volcano early of 2011 after contraction and poro-elastic settling of eruptive deposits 2010 eruption (Kriswati et al ., 2014). (e.g., Lu et al ., 2005; Wadge et al ., 2006; Lee et al ., This study shows how the use of InSAR and time- 2008; Poland, 2010). series technique can facilitate the monitoring of dynamic processes on especially hazardous volcanoes. The spatially dense measurements of surface 4. Discussion and Conclusion deformation help constrain the causes of the observed deformation and the magma plumbing system. The SBAS technique was used to precisely estimate Continuous monitoring of ground deformation at surface deformation rates and time-series Sinabung and other volcanoes worldwide will help measurements at Sinabung volcano between 2007 and reduce volcanic hazards. 2011. Sinabung volcano displayed gradual inflation (~1.7 cm/yr) for approximately 3 years and then transitioned to a period of rapid inflation (~6.6 cm/yr) Acknowledgment at the summit of the volcano during the 6 months prior to the main eruption event. A similarly rapid period of This research was funded by the Korean Society of deflation (~5.8 cm/yr) occurred during the 4 months of Remote Sensing. This research was supported by Basic the post-eruption period. The greatest surface Science Research Program through the National deformation was measured to be 8-10 cm at the summit Research Foundation of Korea (NRF) funded by the of the mountain a few months before the 2010 eruption. Ministry of Science, ICT & Future Planning (NRF- The Sinabung volcano has a relatively simple eruption 2015M1A3A3A02013416) and the Korea cycle model that includes two primary processes. First, Meteorological Administration Research and the magma chamber is fed from below, causing surface Development Program under Grant KMIPA (2015- uplift. The inflation trend is punctuated by episodic 3071). The SAR data used in this study are copyrighted deflation caused by depressurization of the shallow by the Alaska Satellite Facility and Japan Aerospace magma reservoir due to degassing or other processes. Exploration Agency. When the resulting pressure in the magma reservoir is sufficient, magma is forced to erupt through the conduit to the surface until the pressure inside the reservoir References decreases. Second, the cooling of the magma inside the reservoir as well as the thermo-elastic contraction of Berardino, P, G. Fornaro, R. Lanari, and E. Sansoti, the erupted material generated surface deflation as the 2002. A new algorithm for surface deformation 2010 eruption. These observations and interpretations monitoring based on small baseline differential are consistent with previous study of volcanic SAR interferograms, IEEE Transactions on deformation in the area of Sinabung volcano (Chussard Geoscience and Remote Sensing , 40(11): 2375- and Amelung, 2010, 2012). We can compare InSAR 2383. time-series results with GPS observation data from Chaussard, E. and F. Amelung, 2010. Monitoring the

–44 – Monitoring Mount Sinabung in Indonesia Using Multi-Temporal InSAR

ups and downs of Sumatra and Java with D- SAR interferometry, Earth, planets and space , Insar time-series, American Geophysical 60(5): 447-452. Union. Fall Meeting 2010, abstract #G23C- Lee, C.W., Z. Lu, H.S. Jung, J.S. Won, and D. Dzurisin, 0838. 2011. Surface Deformation of Augustine Chaussard, E., and F. Amelung, 2012. Precursory Volcano (Alaska), 1992-2005, from Multiple- inflation of shallow magma reservoirs at west Interferogram Processing Using a Refined Sunda volcanoes detected by InSAR, SBAS InSAR Approach, USGS Professional Geophysical Research Letters , 39(21). Papers , 1769, 453-465. Ebmeier, S. K., J. Biggs, T. A. Mather, G. Wadge, and Lee, C.W., Z. Lu, and H.S. Jung, 2012. Simulation of F. Amelung, 2010. Steady downslope time series surface deformation to validate movement on the western flank of Arenal multi-interferogram InSAR processing volcano, Costa Rica, Geochemistry, technique, International Journal of Remote Geophysics, Geosysttem , 11(12): Q12004. Sensing , 33(22): 7075-7087. Furuta, R., 2009. Case study of small scale surface Lu, Z., Masterlark, T., and Dzurisin, D., 2005. deformation detection using ALOS PALSAR Interferometric Synthetic Aperture Radar differential interferometry, Proc. Fringe 2009 (InSAR) Study of Okmok Volcano, Alaska, workshop , Frascati Italy. 1992-2003: Magma Supply Dynamics and Global Volcanism Program, 2013. Volcanoes of the Post-emplacement Flow Deformation, World, Smithsonian Institution , Journal of Geophysical Research: Solid Earth , http://dx.doi.org/10.5479/si.GVP.VOTW4- 110: B02403. 2013. Lu, Z., J.W. Kim, H. Lee, C.K. Shum, J. Duan, M. Jung, H.S., Z. Lu, and L. Zhang, 2013a. Feasibility of Ibaraki, O. Akyilmaz, and C.H. Read, 2009. along-track displacement measurement from Helmand River Hydrologic Studies Using Sentinel-1 interferometric wide-swath mode. ALOS PALSAR InSAR and ENVISAT IEEE Transactions on Geoscience and Remote Altimetry, Marine Geodesy , 32(3): 320-333. Sensing , 51: 573-578. Lu, Z., Dzurisin, D., Biggs, J., Wicks, C., Jr, and Jung, H.S., D.T. Lee, Z. Lu, and J.S. Won, 2013b. McNutt, S., 2010. Ground surface deformation Ionospheric Correction in SAR Interferogram patterns, magma supply, and magma storage at by Multiple-Aperture Interferometry, IEEE Okmok volcano, Alaska, from InSAR analysis: Transactions on Geoscience and Remote 1. Intereruption deformation, 1997-2008, Sensing , 51(5): 3191-3199. Journal of Geophysical Research: Solid Earth , Kriswati, E., H. Kuncoro, and I. Meilano, 2014. Low 115(B5): B00B02. Rate of Sinabung Deformation Inferred by GPS Lu, Z. and D. Dzurisin, 2014. InSAR Imaging of Measurement, Proc. of AIP Conference , 4th Aleutian Volcanoes , Springer, Berlin International Symposium on Earthquake and Heidelberg. Disaster Mitigation 2014, 1658(1): 050007-1- Massonnet, D. and K.L. Feigl, 1998. Radar 050007-8. interferometry and its application to changes in Lee, C.W., Z. Lu, O. Kwoun, and J. S. Won, 2008. the Earth’s surface, Reviews of Geophysics , Deformation of Augustine volcano, Alaska, 36(4): 441-500. 1992-2005, measured by ERS and ENVISAT Ng, A.H.M., H.C. Chang, L. Ge, C. Rizos, and M.

–45 – Korean Journal of Remote Sensing , Vol .33, No .1, 2017

Omura, 2009. Assessment of radar The central Andes. Kamchatka, and beyond, interferometry performance for ground GSA Today , 14(8): 4-11. subsidence monitoring due to underground Poland, M., 2010. Localized surface disruptions mining, Earth Planets Space , (61): 733-745. observed by InSAR during strong earthquakes Pan, G. and D. Tang, 2010. Damage information in Java and Hawaii, Bulletin of the derived from multi-sensor data of the Seismological Society of America , 100(2): 532- Wenchuan Earthquake of May 2008, 540. International Journal of Remote Sensing , Rignot, E., 2008. Changes in West Antarctic ice stream 31(13): 3509-3519. dynamics observed with ALOS PALSAR data, Philibosian, B. and M. Simon, 2011. A survey of Geophysical Research Letters , 35(12): L1205. volcanic deformation on Java using ALOS Wadge, G., G.S. Mattioli, and R.A. Herd, 2006. Ground PALSAR interferometric time series, deformation at Soufri ére Hills Volcano, Geochemistry, Geophysics, Geosystems , Montserrat during 1998-2000 measured by 12(11): 1-20. radar interferometry and GPS, Journal of Pritchard, M. E. and M. Simons, 2004. Surveying Volcanology and Geothermal Research , volcanic arcs with satellite radar interferometry: 152(1): 157-173.

–46 –